科研之友
微信
新浪微博
Facebook
分享链接
摘要
In recent years advancement in cross field technologies lead the world to the new era of genomic research. Several new technologies have been developed for early detection of critical genetic disease like cancer. Since the conventional morphological and clinical tests are invasive in nature and harmful to human body, researchers are intending to find a noninvasive way to predict cancer associated genes. They find accuracy as a major concern to be taken care of in disease gene identification system. Therefore the authors in this paper made an effort to raise the accuracy level compare to existing technique. Cancer is caused by changes in deoxyribonucleic acid sequence. If those changes can be identified, diseased gene can therefore be recognized exactly. Hence the authors compared the diseased gene with the healthy ones to capture the differences. In order to search such disparity more efficiently and accurately, functional link artificial neural network (FLANN) based adaptive filter with least mean squares algorithm is attempted in the present paper. FLANN filter offers trigonometric expansion of the input sequences which reduces the error to minimum value at faster rate. Here, cancer genes are distinguished from healthy genes based on amount of normalized mean square error, which is estimated through adaptive filter. 131 Genes are used to train the identifier and the proposed technique successfully identifies 369 test dataset. The database is collected from National Center for Biotechnology Information genbank. The performance of the overall system is investigated by measuring sensitivity, specificity and accuracy. The proposed algorithm achieves 86.85% accuracy compared to existing entropy based identifier. Overall system performance of the proposed FLANN based adaptive filter is displayed in receiver operating characteristic curve.
- 出版日期2021-2
